[idea] "Register a machine" for Launchpad.net bugs

[Brett Alton]

I originally had this idea [1] on Brainstorm but then realized that it made no sense having it there.

--------

For bugs related to processor architecture (think the Linux kernel), faulty RAM / amount of RAM, wireless cards (think drivers), etc., I think it would be useful to include information on the PC that is submitting the bug.

Developers would benefit from having certain 'machines' attached to hardware-related bugs because users do not always submit hardware information either due to negligence, ignorance or mistake. This way the bug process could be more efficient by adding information usually collected as `cat /proc/cpuinfo > cpuinfo.txt`, `cat /proc/meminfo > meminfo.txt` or `lspci -v > lspci.txt` without the user using the command line.

Launchpad would gather information about the user's machine by an apt-gettable program (or the user could enter it manually, if need be) and have each machine register to the user.

This way, I could register my workstation, my backup server, my work/web server and some other machines I tinker with to Launchpad.

Then, when creating a bug or experiencing a bug that has already been triaged, I could 'attach' my machine to the bug, giving developers more hardware-related information.

There is of course an obvious privacy issue at stake here, so maybe the registration of machines could be flagged as 'private' or 'public' (if they want to show it off) by the user, with 'private' being the default and developers being the only user who can view 'private' machines.

--------

[1] http://brainstorm.ubuntu.com/idea/14041/

[Diogo Matsubara (matsubara)]

Abel, can you comment on this bug?

[Abel Deuring (adeuring)]

Brett,

I am working on a Launchpad project that is quite similar to what you
are describing. hwtest already collects hardware data (mostly from HAL)
and submits it to Launchpad; the ability to associate bugs with such a
hardware profile is currently missing, but is on the to do list. You
will soon be able to look at hardware data using the Launchpad API.

[Brett Alton (brett-alton-deactivatedaccount)]

That's quite amazing, thank you for your quick response!

I'm a Launchpad tester and love to preform any sort of beta work if you would like me to volunteer.

I can't program in Python however, so I'm of no help in that manner.

Thanks and good luck!

I guess you can close this bug since it wasn't really a bug to begin with.

[William Grant (wgrant)]

I used the HWDB submissions today to track down an elusive arch-specific bug. Having them easily accessible would make things somewhat easier, but it's still mighty useful!

[Abel Deuring (adeuring)]

On 11.10.2008 12:34, William Grant wrote:
> I used the HWDB submissions today to track down an elusive arch-specific
> bug. Having them easily accessible would make things somewhat
> easier, but it's still mighty useful!
>

That's really interesting -- I'd thought that the current ways to access
the submissions would not allow this. You're making me quite curious
(and hearing about current use cases gives me also ideas what should be
done to improve the HWDB): How did you find the interesting submissions,
and which data did you use?

[Brett Alton (brett-alton-deactivatedaccount)]

Abel, I personally would find it useful for kernel bugs with different processors and architectures. I've had troubles installing Linux 2.6.24 (Ubuntu 8.04) as a server AND a desktop yet so work is being done on the bug (or so I see). It's an Intel Celeron 2.4 GHz (quite old), but as an LTS, I would figure this would be top priority as the kernel literally panics and freezes and becomes unusable.

I figured more information on my computer would be welcome to help push this bug further. I'm sure this is happening on other bugs as well.

If machines with their encompassing information were attached to every bug then this welcomed information would be mandatory, but would take no more work than setting it up once and never touching it until you modify your PC or server.

[William Grant (wgrant)]

Abel Deuring wrote:
> On 11.10.2008 12:34, William Grant wrote:
>> I used the HWDB submissions today to track down an elusive arch-specific
>> bug. Having them easily accessible would make things somewhat
>> easier, but it's still mighty useful!
>>
>
> That's really interesting -- I'd thought that the current ways to access
> the submissions would not allow this. You're making me quite curious
> (and hearing about current use cases gives me also ideas what should be
> done to improve the HWDB): How did you find the interesting submissions,
> and which data did you use?

The bug in question was bug #267611. None of us were able to reproduce
it, so it was quite unamusing when it persisted, changed nature, and was
still unreproducable. When I finally got a developer to reproduce it,
the only difference I could see was the arch. I went through all of the
people on the bug that it affected, and their recent Intrepid HWDB
submissions were either amd64 or on hardware that others ran amd64 on.

--
William Grant

[Abel Deuring (adeuring)]

William,

> The bug in question was bug #267611. None of us were able to reproduce
> it, so it was quite unamusing when it persisted, changed nature, and was
> still unreproducable. When I finally got a developer to reproduce it,
> the only difference I could see was the arch. I went through all of the
> people on the bug that it affected, and their recent Intrepid HWDB
> submissions were either amd64 or on hardware that others ran amd64 on.

brilliant idea! That's one thing more to remember for the HWDB -- it
should not be necessray to collect this data manually as you had to do.

[Abel Deuring (adeuring)]

Brett,

> Abel, I personally would find it useful for kernel bugs with different
> processors and architectures. I've had troubles installing Linux 2.6.24
> (Ubuntu 8.04) as a server AND a desktop yet so work is being done on the
> bug (or so I see). It's an Intel Celeron 2.4 GHz (quite old), but as an
> LTS, I would figure this would be top priority as the kernel literally
> panics and freezes and becomes unusable.
>
> I figured more information on my computer would be welcome to help push
> this bug further. I'm sure this is happening on other bugs as well.
>
> If machines with their encompassing information were attached to every
> bug then this welcomed information would be mandatory, but would take no
> more work than setting it up once and never touching it until you modify
> your PC or server.

right, hardware profiles can indeed be helpful. As a first step, we
will provide links between a bug and an HWDB submission; later we might
even allow links between a bug and a specific device (or even
device/driver combination) from a HWDB submission.